Part II – 10 Best Practices to make the most out of your Data Visualization
Posted On 18/01/2017
Data visualization helps highlighting aspects that would usually go unnoticed. It helps speeding up the decision making process. And last but not least it is a great way to help combat growing volume of data. Usually, convincing people is the most difficult part. Starting with these 10 best practice tips, you are already one step closer to convincing them.
Examples of distractions are: 3D charts, over-using grid lines, too much ink in relation to the data, … for the latter, there is a guiding principle: maximize data-to-ink ratio. This means that a 1 – proportion of the chart can be erased without losing meaning. By Edward Tufte’s wise words: “Above all else, show the data.”
7: Do not overuse or misuse color
You should limit the use of color; colors do not replace labels. Humans can perceive millions of different colors, but can only differentiate ± 10.
To maximize the effect, you can color semantically: color values according to their meaning.
Here, there is are 3 different ways to differentiate:
Sequential scale: a continuous scale, the differences are quantitative
Use colors that differ in brightness (e.g.: population density)
Diverging scale: for continuous data with a meaningful “breakpoint”
Use double sequential scale (e.g.: sex ratio: % males vs. females in sub population)
Qualitative / nominal: there are no magnitude differences between the categories.
Use colors that differ in hue (Open VLD, Groen, N-VA)
Also, when you use the same or similar entities over different charts, Color consistently: use the same colour for the same entities everywhere.
8: Highlight only the most important information.
This tip is in the same line with previous one: Colour does draw attention, but we have a limited supply of it. Needless to say that the second image is how not to draw attention.
9: Supply sufficient context for the data.
Make sure your visualizations are self-supporting. Mention the source and your methodology used to compile the raw data. An example of sufficient context is: “results based on a telephone survey with 1000 respondents; calls were made between 2/11/2016 and 6/11/2016. Random dialling was used. Only landlines were called”
Another way to clarify your data is to deliver enough information to put your insight in perspective. For example: “a change in methodology results in a break between 2004 — 2005”
10: Balance nature and importance of information in layout.
Data visualization is somewhat of an art form. This means that composition, aesthetics, … do matter, so pay attention to it! Your Dashboards are read from left-to-right, top-to-bottom. The composition will determine how a human preceptor will read your “story”.
Put these best practices into action!
With these 10 tips, you can craft stunning dashboards and impress your colleagues, clients, partners, etc. Create yours at Cumul.io!